If your company is in a private equity portfolio, sometime in the next 90 days you’re going to get one — possibly two — sales calls that look very different from the usual SaaS pitch. They will be from people who say they’re “from Anthropic’s enterprise services firm” or “from OpenAI’s Deployment Company.” They will offer to embed engineers inside your operations. They will mention that your PE sponsor is on the investor list.
This isn’t a SaaS pitch. It’s the new shape of how the two largest AI labs are going to market — and on May 4, 2026, both of them announced versions of it on the same day.
Here’s what each one actually is, and the four questions a mid-market CFO or CIO needs to answer before the first discovery call.
What Just Happened
Anthropic’s venture — $1.5 billion, announced May 4 with Blackstone, Hellman & Friedman, and Goldman Sachs as founding partners. Anthropic, Blackstone, and H&F each commit $300 million. Goldman commits $150 million. Backing partners include Sequoia, Apollo, GIC, General Atlantic, and Leonard Green. (Blackstone press release, Anthropic news)
The new firm is described as “AI-native” — Anthropic engineers and partnership resources are embedded directly inside the standalone entity. Initial sectors: healthcare, manufacturing, financial services, real estate. Goldman and partners will use their own portfolio companies as the proving ground before opening to other mid-sized businesses, especially in the PE-owned universe. (Bloomberg)
OpenAI’s venture — $10 billion, finalized the same day. Anchored by TPG with 19 investors including Brookfield, Advent, and Bain Capital. (Bloomberg, TNW)
The structure is unusual: a 17.5% guaranteed annual return over five years for the PE consortium. OpenAI commits $500 million up front with an option to add $1.5 billion. The PE consortium contributes ~$4 billion across the five-year window. OpenAI keeps strategic control via super-voting shares; the financial sponsors take the income-oriented economics. Priority sectors: healthcare, logistics, manufacturing, financial services. The delivery model is forward-deployed engineers — explicitly compared to Palantir’s go-to-market.
Both ventures share a structural pattern: embed engineers inside customer orgs, get preferred sales access through PE / asset-manager portfolios. Fortune called the Anthropic version a “shot at the consulting industry.” (Fortune) That framing applies equally to both.
The Four Q3 Procurement Questions
If you’re a CFO, CIO, or COO at a PE-owned mid-market company, the next quarterly board meeting is the right venue to surface these. Each one has a clean answer.
1. Is your PE owner on either investor list?
| Anthropic JV investors | OpenAI Deployment Company investors |
|---|---|
| Blackstone | TPG |
| Hellman & Friedman | Brookfield |
| Goldman Sachs | Advent International |
| Sequoia Capital | Bain Capital |
| Apollo Global Management | (15 additional investors not yet fully disclosed) |
| GIC | |
| General Atlantic | |
| Leonard Green |
Pull your cap table. If your sponsor or holding company is on either list, you’ll get a preferred-channel pitch in the next 90 days. If you’re owned by a Blackstone, H&F, Goldman, Sequoia, Apollo, GIC, General Atlantic, or Leonard Green portfolio company, the Anthropic services firm will reach out. If TPG, Brookfield, Bain, or Advent, the Deployment Company will. If both — your sponsor is sophisticated enough to have both bets — you’ll get pitched twice.
The structural fact: the deal is not market-priced. Pricing through these channels will be different from what a non-portfolio-company would pay. Your sponsor will frame this as a benefit. The frame is mostly correct — but verify the actual terms before you treat it as one.
2. Does the embedded-engineering model fit your delivery org?
This is the question that gets skipped most often. Anthropic engineers (or OpenAI engineers) landing inside your operations is closer to a Big-4 consulting engagement than a typical SaaS rollout. They need:
- Access to your production data and operational workflows
- A clear scope and a named internal sponsor at your end
- A change-management runway (your line ops will need to absorb new ways of working)
- Legal review of the embedded-engineer contract — IP ownership, data residency, exit clauses
If your shop has never run a Big-4 engagement, this model is a step-change in how you absorb external resources. If you have, the question is whether your delivery org has the bandwidth for another one. The honest answer for many sub-200-person shops is “not this quarter.”
3. What’s the conflict-of-interest read with your existing AI consulting?
Both ventures explicitly compete with the consulting industry. (Fortune’s framing) If you have an active engagement with Accenture, Deloitte, McKinsey, or BCG on AI workflows, the new venture will pitch to displace it. The pitch will be:
- “We’re closer to the model — direct engineering access”
- “Our pricing is outcomes-aligned, not staffed-rate”
- “We embed for 6-12 months versus a 90-day discovery”
Some of those claims will be true. Some will be marketing. The risk you’re managing is whether mid-engagement vendor displacement makes sense — usually it doesn’t, but the new offers will reset what “best price for AI services” looks like for your renewal cycle.
4. What’s your model-pin lock-in story?
This is the single most consequential question. A 3-year embedded-engineering engagement on Claude effectively pins your model choice to Anthropic for the engagement term. Same for OpenAI on the other side. The “switch models if a better one ships” optionality you have today disappears under either contract.
For some shops, model-pin doesn’t matter — the workflow value comes from the embedded team, not the model identity. For shops where you’ve explicitly built model-portability into your AI strategy (because you watched the Mistral / DeepSeek / Llama curve), an embedded-engineering JV is structurally counter to that strategy.
The exit story matters: what happens at month 36 if you don’t renew? Who owns the workflows the engineers built? Who maintains them? These belong in the contract before the kick-off call, not after.
What This Means for You
If you’re a CFO at a PE-owned mid-market company: Surface these four questions at the next quarterly board meeting. The conversation belongs at board-level, not procurement-level — because the PE sponsor has signaled the channel preference and the COO needs to weigh in on delivery readiness.
If you’re a CIO at the same company: The model-pin question is yours to answer before procurement runs the RFP. Make sure the board hears your read on whether a 3-year Claude-pinned or GPT-pinned engagement fits the architecture you’ve been building.
If you’re a COO running operations: The embedded-engineering question is the one to push back on if you’re not staffed for it. Your delivery org carries the absorption cost. A no-this-quarter from you is a legitimate answer.
If you’re at a non-PE-owned mid-market company (founder-owned, family-owned, or public): The preferred-channel pitch isn’t aimed at you. You’ll see open-RFP versions of both offerings within 6-12 months as the ventures expand beyond their PE-portfolio proving ground. Wait — the PE-channel customers will surface the bugs first.
If you’re at a Big-4 / consulting firm: The displacement risk is real. The two largest customers of mid-market AI consulting (Anthropic and OpenAI accounts) are now your direct competitors with structural advantages — model access, embedded-engineer model, sponsor relationships. The defense is the multi-vendor argument and the deeper change-management capability.
If you’re a journalist or analyst writing about this: The 17.5% guaranteed return on the OpenAI side is the unusual structural detail worth pulling on. Standard PE returns are equity-aligned, not income-aligned. This deal is closer to a structured-finance product than a typical venture investment — which suggests OpenAI is buying a distribution channel using its balance sheet rather than equity dilution.
What This Doesn’t Cover
Pricing transparency. Neither venture has published pricing. The investor framing suggests outcomes-or-engagement-basis billing, but the actual invoicing structure is not yet public. Treat the eventual quote as the first piece of real data.
Sub-100-person shops. Both ventures target mid-market — Bloomberg’s coverage uses “mid-sized” repeatedly. If your company is sub-100 employees, the embedded-engineering overhead is probably structurally too high. The same lab capability is available through normal API access at a fraction of the operational cost.
Multi-cloud / hybrid model strategies. If your AI architecture explicitly includes Bedrock-routed Claude, Vertex-routed Gemini, and direct OpenAI API as a portfolio, neither JV fits without compromise. The JV model assumes single-vendor depth.
Regulated-industry restrictions. Healthcare, financial services, and government workloads have data-residency, audit-trail, and vendor-management requirements that may bar embedded-engineer arrangements outright. Talk to legal before the discovery call, not after.
The Big-4 response. Accenture, Deloitte, McKinsey, BCG haven’t publicly responded yet — but the displacement pressure is structural. Their counter-pitch will land within 6 months. The intervening period is when CFOs have the most negotiating leverage on both sides.
The Bottom Line
The same-day twin-launch isn’t a coincidence. It’s the two largest AI labs telling the market that frontier-lab capability without distribution is no longer the bet — distribution through PE / asset-manager portfolios is. For mid-market CFOs and CIOs at PE-owned companies, that translates into a procurement decision that’s bigger than picking a SaaS vendor.
The four questions above answer it. The Q3 quarterly meeting is the venue to ask them. Your sponsor relationship is the angle that makes the conversation different from a normal RFP.
For a foundation in how to think about agent-based business automation across your stack — not just one vendor’s flavor — our AI Business Automation course covers the architectural decisions that pre-date the vendor choice. AI Agents Deep Dive is the technical companion for the engineers who’ll be evaluating the embedded-team offers.
Surface the four questions. Your sponsor knows they’re coming. Your board doesn’t yet.
Sources
- Anthropic Partners with Blackstone, Hellman & Friedman, and Goldman Sachs to Launch Enterprise AI Services Firm — Blackstone press release
- Building a new enterprise AI services company — Anthropic
- Anthropic and OpenAI are both launching joint ventures for enterprise AI services — TechCrunch
- Anthropic teams with Goldman, Blackstone and others on $1.5 billion AI venture targeting PE-owned firms — CNBC
- Goldman, Blackstone Partner With Anthropic on AI Services Firm — Bloomberg
- Anthropic takes shot at consulting industry — Fortune
- OpenAI Finalizes $10 Billion Joint Venture With PE Firms to Deploy AI — Bloomberg
- OpenAI closes The Deployment Company, a $10bn enterprise AI bet on private equity — TNW
- OpenAI Guarantees 17.5% Annual Return in Unusual $10 Billion Enterprise Deal — Technobezz
